842 



Fishery Bulletin 101(4) 



fected by sample size (Magurran, 1988), was marginally 

 nonsignificant (HB, treatment, P=0.07, Table 3). 



cies diversity (HB), though not richness (S), was signifi- 

 cantly affected by sample date (P=0.02). 



Presence or absence of a fouling community 



Treatment and sample date combined to have a significant 

 efTect on the number of A. troschelii (date by treatment 

 interaction, P<0.01) — an effect contributing to the similar 

 significant interaction effect for all species combined (Fig. 

 7, Table 4). Although the mean numbers of fish(es) were 

 consistently higher at fouled FADs i'or E. bipinnulata alone 

 and for all species minus A. troschelii, the only significant 

 main efTects were due to sample date (Table 4, Fig. 7). Spe- 



Artificial fish experiment 



Experimental treatments (FADs with model fish, with lead 

 weights or with nothing, Fig. 4) had no effect on any mea- 

 sured parameter — combined species, A. troschelii alone, E. 

 bipinnulata alone, combined species less A. troschelii, spe- 

 cies richness and diversity (Fig. 8, Table 5). All measures 

 were significantly affected (P<0.01) by sample date except 

 for£. bipinnulata alone (date, P=0. 48). Although individual 

 FADs varied in the number of associated E. bipinnulata. 



